Datasets Skill - Discovering system and user-defined datasets

Dataset categories

Use list_datasets to discover available datasets:

• System datasets: source system/<dataset>
• User-defined datasets: source default/<dataset>

Core workflow

1. Call list_datasets first.
2. Use the exact source ... path returned by list_datasets. Never guess a dataset name.
3. For any system/... dataset, always call get_system_dataset_info to retrieve dataset information and schema guidance before querying it.
4. Load or use coralogix-dataprime skill to build the actual query correctly.
5. If the user's question should be answered from a specific dataset, call query_dataprime with that dataset in the source ... clause.

When to use get_system_dataset_info

Use get_system_dataset_info only for system/... datasets, and always use it after discovery before querying a system dataset.

It is the required follow-up step for system datasets because it is intended to provide:

• what a specific system dataset is for
• whether a specific system dataset should be queried
• schema-oriented information for a specific system dataset

Do not use it for default/... datasets. In this scope, there is no matching detailed-info tool for user-defined datasets yet.

Important notes

• search_for_fields is only for logs and spans, not arbitrary datasets.
• For a known system dataset with dedicated business logic or schema guidance, load the relevant domain skill as well.
• For fleet case analytics on system/labs.cases.state_updates, also load coralogix-case-analytics.